The original study indicated that the preventive effect of soy on breast cancer in premenopausal women, if it exists, is not mediated by circulating sex hormones. The proposed study examines 2 alternate hormonal mechanisms of action for soy: modification of estrogen levels and epithelial cells directly in the breast and modulation of urinary estrogen metabolite patterns. The objectives are to examine the effects of soy on estrogen levels in nipple aspirate fluid (NAF), on cytologie patterns of epithelial breast cells in NAF, and on estrogen metabolism as expressed in the formation of urinary 2,16a, and 4-hydroxy metabolites. In addition, estrogen levels in NAF and serum will be compared. In a crossover design, we will randomize 100 women aged 30-45 years recruited through a health maintenance organization into 2 groups. The goal is to obtain sufficient NAF for hormone and cytologie analysis from 80 women. Group A will first undergo 6 months of a high soy diet and then, after a 3-month washout period, switch to 6 months of a low soy diet. Group B will follow the 2 diets in reverse order. The high soy diet will consist of 2 daily servings of tofu, soymilk, or soynuts;the low soy diet will contain less than 3 weekly soy servings. Dietary counseling, gift certificates, and soy foods will help to maintain adherence. Compliance will be assessed by urinary isoflavone excretion, 24-hour dietary recalls, and soy logs. Both groups will donate NAF and urine at baseline and then every 2 months. Blood for estrogens will be collected at baseline and at the end of each diet period. Part of the NAF samples will be diluted in buffer and frozen. The other part will be fixed on cytoslides for evaluation according to defined morphologic criteria. Estradiol and estrone will be measured in NAF and serum with an ultrasensitive radioimmunoassay, urinary estrogen metabolites by GC-MS, and urinary isoflavones by LC- MS. The statistical analysis will apply general linear models to compare hormone levels, cytology of breast cells, and urinary estrogen metabolites by soy intake, while taking into account the repeated measurements. This project is relevant to elucidate the role of soy in breast cancer development and prevention. Despite significant evidence for a protective role of soy, it is not understood how soy may influence breast cancer risk. The NAF approach is the only way to learn more about the beneficial and adverse effects of soy directly in the female breast without applying invasive procedures to women without breast disease or surgery.